Network optimization problems are modelled as graphs, with nodes and weighted edges representing flow, cost, or distance.
Flow Problems
LP formulations of these models are covered in linear programming.
- Transportation problem
Minimizes the cost of shipping goods from a set of supply nodes to a set of demand nodes. - Transshipment problem
Generalizes the transportation problem by allowing goods to pass through intermediate nodes. - Assignment problem
Minimizes the cost of assigning agents to tasks, 1 agent per task.
Network Algorithms
- Minimum connector problem
Finds a minimum weight tree connecting all nodes, solved by algorithms such as Kruskal or Prim. - Maximum flow problem
Finds the maximum flow from a source node to a sink node subject to edge capacities. - Shortest and longest path algorithms
Finds the minimum or maximum weight path between 2 nodes.
Engineering Project Management
Critical Path Analysis
Finds the longest path through a project’s task network, determining minimum project duration.
- Forward pass
Computes each task’s earliest start and earliest finish from project start. - Backward pass
Computes each task’s latest start and latest finish from project end. - Float
Latest start minus earliest start. The amount a task can slip without delaying the project. - Critical path
The chain of tasks with 0 float. Its total duration is the minimum project duration.
PERT Model
Estimates task durations probabilistically using optimistic, pessimistic, and most likely estimates.
Here:
- : optimistic estimate
- : most likely estimate
- : pessimistic estimate
- : expected task duration
Schedule Adjustment
- Project crashing
Reduces project duration by allocating extra resources to critical tasks at extra cost. - Resource smoothing
Reschedules tasks within their float to balance resource usage, without extending project duration. - Resource leveling
Reschedules tasks to keep resource usage under a fixed limit, extending project duration if necessary.